What it does
An end-to-end pipeline: speech to text, scripts written against timestamps, a cloned English voice, Tamil text to speech, and automatic retiming of the video so every line lands on its scene.
The problem
Dubbing a long video by hand means transcribing, translating, recording and editing every scene. Machine translation alone drifts out of sync and reads badly when spoken.
Approach
Scripts are written per timestamped block, so each line has a time budget. The English and Tamil voices come from AI voice models, chosen after testing four voice engines. When a line runs long or short, the video block is stretched or trimmed instead of rushing the voice.
- Input
- AI step
- Rules or code
- Person decides
- Result
Screens
Swipe sideways to see the whole screen.
Client work is shown under confidentiality agreements: no client names, and screens are recreated with sample data.
Stack
- Gemini API
- AI voice models
- Video editing
Impact
Typical for systems like this.
